1. Field of the Invention
The present invention relates to a micromechanical device using surface micromachine technologies.
2. Description of the Related Art
Electrically controlled switching elements used in various electronic devices include semiconductor (solid-state) switches and reed relays. From the standpoint of an ideal relay, they have merits and demerits.
The semiconductor switches have merits of being capable of being downsized and operating at high speed and being high in reliability. They can also be easily integrated as an array of switches. For instance, PIN diodes, HEMTs (High Electron Mobility Transistors) and MOSFETs have been used as switches for switching antennas adapted for microwave, millimeter wave, etc. In comparison with switches which close or break mechanical contacts, however, the semiconductor switches are high in the on impedance and low in the off impedance and have large stray capacitance.
In comparison with the semiconductor switches, on the other hand, the reed relays are high in the on/off impedance ratio and can be designed to minimize insertion loss and ensure signal fidelity. For this reason, the reed relays have been frequently used in semiconductor testers by way of example. However, they are large in size and low in switching speed.
Recently, attention has been paid to micromechanical switches which have the merits of semiconductor switches and reed relays. Among others, micromechanical switches that are formed using surface micromachine technologies and are operated electrostatically can be implemented at low cost because they can be formed through the use of semiconductor thin-film techniques.
FIG. 1A is a plan view of a conventionally proposed micromechanical switch and FIG. 1B is a sectional view taken along line 1B—1B of FIG. 1A. This switch has a source electrode 51, a drain electrode 52, and a gate electrode 53 therebetween, which are all formed on a substrate 50 made of, say, silicon. A conductor beam 54 is formed above the gate electrode 53 with a predetermined gap therebetween. Although the electrodes are named source, drain and gate after those of MOSFETs, the switch is different in structure from the MOSFETs.
The conductor beam 54 has its one end fixed to the source electrode 51 to form an anchor portion 55. The other end of the beam is made open to form a moving contact (contact chip) 56. When a voltage is applied to the gate electrode 53, the conductor beam 54 is deflected downward by resulting electrostatic force, allowing the moving contact 56 to come into contact with the drain electrode 52. When the gate electrode 53 is deenergized, the conductor beam 54 is restored to its original position.
An analysis of deflection of the conductor beam using a mechanical model has been made by P. M. Zavracky et al. (“Micromechanical Switches Fabricated Using Nickel Surface Micromachining” Journal of Microelectromechanical Systems, Vol. 6, No. 1, March 1997). According to this analysis, when gate voltage is applied, the conductor beam 54 connected to the source electrode 51 is held in a position d(x) above the gate electrode 53 with x as the distance from the source. The gate voltage required to hold the conductor beam 54 in a deflected state increases monotonously with increasing deflection but, after it has been deflected to a certain extent or more, decreases monotonously. The system therefore becomes unstable. At some gate voltage (threshold voltage Vth), the beam bends, closing the switch.
The threshold voltage Vth according to this model is represented byVth=(2/3)×d0×(2kd0/3 ∈0A)1/2where d0 is the initial gap between the conductor beam and the gate electrode, k is the effective spring constant of the conductor beam, A is the area of portions of the conductor beam and the gate electrode which are opposed to each other, and ∈0 is the dielectric constant of air.
From this it can be seen that Vth is lowered by increasing A (increasing electrostatic force acting on the beam), reducing k, and decreasing d0. However, reducing k results in a reduction in maximum switching speed and decreasing d0 results in an increase in electrostatic coupling between the gate electrode and the conductor beam. Another method of lowering Vth is to increase the amount of downward projection of the moving contact 56, i.e., to decrease the gap g between the moving contact 56 and the drain electrode 52. Thereby, the switch can be closed before the unstable point is reached.
Thus, manufacturing of the gaps d0 and g with precision is essential in lowering the threshold voltage Vth. The manufacture of the micromechanical switch involves complicated processes. To be specific, the source electrode 51, the drain electrode 52 and the gate electrode 53 are first formed on the substrate. A sacrificial layer of, say, silicon oxide, is then deposited on these electrodes. The sacrificial layer is subjected to two-step etch processing. In the first step, the sacrificial layer is partly etched to form the contact chip portion 56. In the second step, in order to form the anchor portion 55, the sacrificial layer is etched until the source electrode 51 is reached.
Subsequently, a conductive layer is deposited over the sacrificial layer and then patterned. Finally, the sacrificial layer is etched away in order to separate the conductor beam 54 from the substrate.
To manufacture the micromechanical switch as described above, the following four lithographic processes (masking processes) are involved:
(1) Patterning of the source electrode, etc.
(2) Patterning of the contact chip portion in the sacrificial layer
(3) Patterning of the anchor portion in the sacrificial layer
(4) Patterning of the conductive layer
There has also been a proposal for use of a mechanical vibrator manufactured through similar micromachine technologies as a high-frequency filter; in fact, a bandpass filter of the order of 100 MHz has been manufactured (see C, Nguyen, et at., “VHF free-free beam high Q michromechanical resonators.” Technical digest, 12th International IEEE Micro Electro Mechanical Systems Conference, 1999, pp. 453–458). The advantages of mechanical vibrator filters are that the Q value is extremely high in comparison with electrical LC filters and the size can be made extremely small in comparison with dielectric filters and SAW filters.
FIG. 2A is a plan view showing the unit configuration of such a vibrator filter and FIG. 2B is a sectional view taken along line 2B—2B of FIG. 2A. A vibrator 61, an input terminal 62 and an output terminal 63 are formed on a substrate 60 by means of micromachine technologies. The vibrator 61 is formed of polycrystalline silicon integrally with four supporting beams 64a to 64d. The supporting beams 64a to 64d have their ends fixed to the anchors 65a, 65b, and 65c, whereby the vibrator 61 is held floating above the substrate.
As with the vibrator 61, the input terminal 62 is formed from a film of polycrystalline silicon. The underlying metal is extended so that its one end is located just below the vibrator, forming a gate electrode (driving electrode) 66. The output terminal 63 and the vibrator 61 are formed on a common metal electrode 67. In practice, a mechanical filter with a given passband is manufactured by connecting a plurality of such unit vibrator filters in parallel with one another.
The vibrator 61 is driven by the driving electrode 66 to vibrate in an up-and-down direction. The resonant frequency f0 of the vibrator 61 is represented by f0=(½π) T(k/m)1/2 where k is the spring constant of the vibrator and m is the mass of the vibrator. With the structure and dimensions in FIGS. 2A and 2B, since k=3Eh3b/l3 and m=ρLwh, f0=(1/π) (Eh2b/ρLwl3)1/2 where E is the young's modulus of the vibrator and ρ is the density.
For silicon, E=1.7×1011 Pa and ρ=2.33×103 kgm−3.
In a typical case with L=13.1 μm, l=10.4 μm, w=6 μm, h=2 μm and b=1 μm, f0=92 MHz.
With portable terminals, use is made of a frequency band of 800 MHz to 5 GHz. For such applications, it is desirable to use mechanical filters which are adapted for higher frequencies than conventional ones. FIG. 3 shows the configuration of such a high-frequency receiver, which includes a bandpass filter 171, a low-noise amplifier 172, a bandpass filter 173, and a mixer 174. The mixer 174 is controlled by a phase control circuit 175 having a PLL (Phase-Locked Loop)/VCO (Voltage Controlled Oscillator). It is also desirable to use mechanical filters for the bandpass filters 171 and 173 and the PLL/VCO in the phase control circuit 175.
In order to implement a high-frequency version of the filter shown in FIGS. 2A and 2B, one might suggest increasing h, increasing b and/or decreasing L and/or l. However, this is not easy with current semiconductor processes. The structure and processes are also complicated.
As described above, the micromechanical switch proposed so far is complicated in manufacturing process and difficult to lower the threshold voltage. In particular, it is difficult to lower the threshold voltage because the contact-to-contact spacing (gap) g depends on the thickness of the sacrificial layer and the amount by which it is etched. The conventional micromechanical vibrator is also complicated in both structure and process and difficult to make a high-frequency version thereof.
For this reason, there has been a demand for a micromechanical device which is allowed to have a high performance characteristic with simple structure and a method of manufacture thereof.